Atmospheric particles represent a large source of uncertainty in global radiative modeling, whichis driven largely by an incomplete understanding of the processes responsible for their formation and growth. Particles smaller than 100 nm in diameter, termed ultrafine particles, are of particular interest because of their potential to grow into cloud condensation nuclei (CCN). Their ability to act as CCN depends on numerous factors, including their chemical composition and physical properties. These properties depend on the conditions under which the particles formed, meaning that an improved understanding of these conditions will improve global climate modeling. There is currently a poor understanding of the processes leading to the formation and growth of these small particles in remote locations. This dissertation investigates the processes that lead to ultrafine particles in two remote environments: the Alaskan Arctic and the Amazon Basin.

In Chapter 2, we reported indirect measurements of ultrafine particle composition made duringMarch 2009 in Utqiagvik, Alaska. We compare measurements of ambient size-selected ultrafine particles and those measured during two ultrafine particle growth events, all of which occurred during periods with minimal local emissions. The ambient particles were found to be moderately hygroscopic, with measured hygroscopic growth factors (HGFs) ranging from 1.45 to 1.51 at 90 % RH, and largely volatile. Combining these data, we estimated that the volume of these particles were comprised of oxidized organics (∼70 %) and ammoniated sulfates (∼30 %). The first ultrafine particle growth event was associated with both solar radiation and elevated levels of sulfuric acid at the site, and analysis of air mass back-trajectories indicated that this event was influenced by the upper marine boundary layer above the Arctic Ocean. The second event was not associated with solar radiation or sulfuric acid. Air masses for this event were close to the surface of the Arctic Ocean and passed over open leads, suggesting influence from marine emissions on the observed particle composition. Particles in Event 1 (HGF = 2.1 for 35 nm particles) were more hygroscopic than those in Event 2 (HGF = 1.67 for 15 nm, 1.94 for 35 nm), though both were similarly volatile. From these data we estimated that particles in both events were largely comprised of a highly-hygroscopic and volatile sea salt-like compound (Event 1: 74 %; Event 2: 15 nm - 63 %, 35 nm - 74 %), with particulate volume in Event 1 balanced by sulfuric acid (22 %) and oxidized organics (4 %) and that in Event 2 balanced by a large fraction of oxidized organics (15 nm: 37 %; 35 nm: 26%).

The next two chapters of this thesis focus on measurements from the Amazon Basin. Chapter3 reports measurements of sulfuric acid from the Amazon Basin made during the Green Ocean Amazon (GoAmazon2014/5) field experiment during both the wet and dry seasons. There was little difference in median concentrations measured between the wet (7.82x105 molec cm−3) and dry seasons (2.59x105 molec cm−3). While these measurements are consistent with those made in Hyytiälä, Finland, unlike in Hyytiälä there was no obvious correlation between sulfuric acid and radiation. Additionally, we evaluated the ability of existing sulfuric acid parameterizations, which estimate sulfuric acid concentrations based on the concentrations of its sources and sinks, to estimate measurements from the Amazon Basin. None of the parameterizations effectively estimated nighttime measurements. We hypothesize that nighttime sulfuric acid is produced through both oxidation of sulfur dioxide by hydroxyl radical and a stabilized Criegee intermediate pathway; the former is not included in any current proxy. Several existing proxies are effective for daytime estimates.

Chapter 4 reports measurements of ultrafine particle number-size distribution from the TapajósNational Forest, an eastern Amazonian rainforest, made during the transition from the wet-to-dry season (May - July, 2014). Frequent bursts of ultrafine particles were observed at the measurement site. Ultrafine particle growth events decreased in frequency from May through July, as did the number of particles 5 - 30 nm in diameter. The concentration of particles in this size range was correlated with boundary layer height in May, suggesting that particle formation during the wet- ter times of year occurs from nucleation aloft or atmospheric dilution. This correlation decreases through June and July, indicating that other particle formation mechanism contribute during the drier times of year. Air mass back-trajectories suggest that aged anthropogenic influence becomes stronger at the site throughout this transition, which is supported by the observed increase in particles 100 - 110 nm

Atmospheric nanoparticle formation and growth processes are major sources of uncertainty in our understanding of global climate. However, nanoparticle composition is notoriously difficult to measure below 30~nm due to their incredibly low mass, and so a full understanding of the compounds that contribute to nanoparticle growth still remains elusive. In addition, nanoparticle physical and chemical properties are continuously changing at different sizes. To date, nanoparticles have been known to form from either the condensation of extremely low volatility organic compounds or the reactions of acids and bases in the atmosphere. For the latter, sulfuric acid has been observed to be the predictor of new particle formation events and, recently, amines and ammonia have been shown to stabilize sulfuric acid-containing clusters and contribute to nanoparticle formation and growth. However, size-resolved composition of nanoparticles resulting from these acid-base pairs, as well as the extent to which other atmospherically relevant acids and bases contribute to new particle formation, are both still unknown. This dissertation uses both modeling and experimental methods to understand different acid-base systems’ nanoparticle formation and growth processes.

In Chapter 2, we generated particles made from sulfuric acid with either dimethylamine and ammonia under dry and humid conditions to measure their size-resolved composition using Thermal Desorption Chemical Ionization Mass Spectrometry. Sulfuric acid-ammonia experiments were conducted with 100x higher concentrations of ammonia compared to sulfuric acid to mimic atmospheric conditions, while sulfuric acid-dimethylamine experiments were conducted with concentrations on the same order of magnitude. In all cases, the particles that were formed deviated from stoichiometric neutrality. In both humid and dry cases, sulfuric acid and ammonia formed particles that contained more acid than base, despite the starting concentration differences. In humid conditions, sulfuric acid and dimethylamine formed particles that also had a higher concentration of sulfuric acid compared to dimethylamine for particles smaller than 12 nm. However, under dry conditions, these particles seemed to have more dimethylamine than sulfuric acid. Thermodynamic models using different pK\textsubscript{a} values to describe the acid and base behavior in these systems are discussed.

Next, in Chapter 3, we probed the composition of particles formed from nitric acid and dimethylamine using the same methods. In this case, under all conditions, nitric acid and dimethylamine formed particles in a 1$:$1 acid:base ratio at all sizes measured (9--30 nm). Here we also applied quantum chemical methods to calculate cluster stabilities for clusters containing up to 4 acid and 4 base molecules to discern if the initial growth pathways were reflected in composition measurements. Results show that cluster simulations predicted the behavior of nitric acid and dimethylamine particles, because both nitric acid and dimethylamine are too volatile to remain in the particle phase without salt formation.

In order to better understand the very first steps of new particle formation, we used the same computational theory to model and calculate the stabilities of acid-base heterodimers (1 acid and 1 base clusters) formed from 3 different acid and 9 different base molecules in Chapter 4. We compared these values to aqueous-phase acidity, gas-phase acidity, base vapor pressure, dipole moment, and polarizability to find out which was the strongest predictor of heterodimer stability, the first step in forming a cluster. Then, we compared heterodimer stability of just the sulfuric acid salts with calculated new particle formation rates. From this, we developed a model that parametrizes heterodimer stability to predict new particle formation rates, which were found to be in good agreement with experimental values for the sulfuric acid-ammonia system.

Accurately representing the sources of important precursor gases for new particle formation and their chemical reactions in the atmosphere are critical for developing a robust understanding of the impacts of ultrafine particles on climate and health. Many types of compounds contribute to these nucleation and growth processes, including inorganic acids, bases and low-volatility organic compounds. Both formation of these compounds in the atmosphere and the processes by which they are converted to ultrafine particles depend greatly on their composition. Therefore, in this dissertation, these sources and processes were investigated using measurements of chemical composition, specifically chemical ionization mass spectrometry (CIMS). CIMS was used to measure important precursor gases for ultrafine particle formation and growth in controlled laboratory chamber studies for three separate chemical systems.

In Chapter 2, the chemical composition of gasses emitted from vehicle brake wear were measured using both nitrate reagent ion (nitrate-CIMS) and iodide reagent ion (iodide-CIMS). Gases, coarse-mode particles and ultrafine particles were generated from a home-built brake dynamometer system, which simulated moderate and heavy braking conditions. The brake pad formulation (ceramic vs semi-metallic), degree of wear (new, gently worn, heavily worn), and temperature of the brake rotor influenced measured gaseous emissions. A critical temperature range was observed for ultrafine particle formation. These critical temperatures were found to be lower for ceramic brake pads (100-140 C) than semi-metallic brake pads (140-200 C), and increase as the brake pads became worn. High concentrations of sulfuric acid and sulfur trioxide were generated during heavy braking conditions at high temperatures, and were more abundantly emitted by the semi-metallic pads than the ceramic. Moderate braking conditions produced NOx, CO, inorganic nitrogen compounds HONO, N2O5, oxidized organics and organic acids. Emissions from different brake pad formulations and degrees of wear were highly variable, emphasizing the importance of representing brake-wear emissions over the lifetime of a brake pad for future control strategies.

In Chapter 3, the effects of monoterpene structure on nitrate radical-induced oxidation mechanism of four monoterpenes (a-pinene, b-pinene, d-carene, a-thujene) was investigated by conducting laboratory chamber experiments. Oxidized reaction products were measured with nitrate-CIMS. The major gas-phase species produced in each system were distinctly different, showing the effect of monoterpene structure on the oxidation mechanism and further elucidated the contributions of these species to particle formation and growth. Measured effective O:C ratio of reaction products were anti-correlated to new particle formation intensity and number concentration for each system; however, monomer:dimer ratio of products had a small positive trend. Gas phase yields of oxidation products correlated with particle number concentrations for each monoterpene system, with the exception of a-thujene, which produced a considerable amount of low volatility products but no particles. The lack of measurable new particle formation in spite of the presence of these dimers indicates a more complex relationship between highly oxidized organic compounds and new particle formation.

In Chapter 4, the formation of low-volatility gasses and new particle formation from ozonolysis of a-thujene was observed in laboratory chamber studies. This system was found to produce lower yields of particles at all mass loadings compared to s-pinene ozonolysis systems. The unique structure of a-thujene led to the competitive production of semi-volatile fragment compounds that reduced the overall volatility of the oxidized products. Theoretical growth rates (on the order of nm per min) calculated from volatility-parameterized organics underpredicted measured growth rates. However, the simple condensational growth model that was used did not account for the irreversible partitioning of semi-volatile organics, which were abundantly produced by this system.